CEOInterviews.AI
Start App
Try: AI strategy tariffs earnings guidance layoffs interest rates growth

Liam Fedus

Co-founder; formerly VP of Research, OpenAI, Periodic Labs

Search every verified Liam Fedus interview, podcast appearance, and on-the-record quote — each transcript cross-checked by AI and human review to confirm speaker identity. Liam Fedus, co-founder and co-CEO of Periodic Labs, has been publicly discussing the company's work on building AI systems for scientific discovery, which he describes as "synthesis superintelligence." In recent appearances, he stated that Periodic Labs is constructing autonomous physical laboratories to generate experimental data, arguing that scientific discovery requires real-world experimentation beyond what language models trained on existing text can achieve. He said that machine learning performs well on trained tasks but that discovery involves untrained territory, and he described the company's approach as training AI on the scientific process itself, including noisy data interpretation and decision-making under uncertainty, rather than only on finished papers or textbooks. Fedus also discussed his background, noting that he helped create ChatGPT and later led post-training as VP of research at OpenAI before leaving to co-found Periodic Labs. In a keynote, he mentioned that the company trained a model called Neon using 1,300 H200s, which he characterized as a relatively small amount of compute compared to frontier labs, and he highlighted infrastructure innovations built on open-source software. He expressed views on the potential impact of AI on science, saying that successful efforts could increase the number of scientists and accelerate progress in fields like materials science, and he emphasized the importance of recording detailed experimental context to improve model learning.

Recent appearances

  • AI Scientists Are Here: Autonomous Labs & Synthesis Superintelligence — Periodic Labs

    October 07, 2026 · 📺 Latent Space

    From building AI systems that reason over noisy physical experiments to creating laboratories where every instrument can become intelligent, Periodic Labs is betting that the next frontier of AI won’t come from simply training on more internet data — it will come from letting models experiment with the real world. In this episode, Periodic Labs’ Liam Fedus and Ekin Dogus Cubuk join swyx and Brandon to explain why scientific discovery is fundamentally different from math and coding, and what it takes to build AI scientists that can actually discover new materials. We go deep on Periodic’s visi…

    Watch on YouTube ↗

  • Liam Fedus - Co-Founder of Periodic Labs

    October 01, 2026 · 📺 Mercor

    Liam Fedus helped build ChatGPT and later led post-training as VP at OpenAI before leaving to co-found Periodic Labs, which is trying to build an AI scientist. His argument is that language models had the internet to learn from, but science has no equivalent, so Periodic had to build physical labs to generate that data itself. He and Brendan Foody talk through what that has meant in practice, from the first months after ChatGPT's launch to a small lab that got the company stuck faster than they expected. In this conversation: The chat interface almost lost to a meeting-bot pitch that would ha…

    Watch on YouTube ↗

  • Meet the Ex-OpenAI & DeepMind Leads Creating an AI Scientist

    September 29, 2026 · 📺 The Generalist

    Liam Fedus and Dogus Cubuk are the co-founders and co-CEOs of Periodic Labs, a startup building “synthesis superintelligence.” In practice, that means an AI system that closes the loop between hypothesis, experiment, and learning. Liam previously led post-training at OpenAI, where he was one of the creators of ChatGPT. Dogus spent years at Google DeepMind leading a large team of chemists and materials scientists. Together, they’ve brought those two trajectories to bear on a problem that has stymied science for decades: not just discovering new materials, but making them reliably and at scale…

    Watch on YouTube ↗

  • Keynote: AI Scientists: Automating Scientific Discovery | Day 2

    September 28, 2026 · 📺 Fellows Forum

    Keynote: AI Scientists: Automating Scientific Discovery Speaker: Liam Fedus, Co-founder at Periodic Labs Recorded at Fellows Forum 2026 (September 23 to 24, 2026). Fellows Forum: https://fellowsforum.com Fellows Fund: https://fellowsfundvc.com

    Watch on YouTube ↗

  • Training AI on the Scientific Process | Liam Fedus (Periodic Labs) | Ray Summit 2026

    September 17, 2026 · 📺 Anyscale

    "Today's models are trained on the finished record of science: the final paper, the final textbook. Periodic Labs is training AI on the scientific process itself, with autonomous physical labs as its reinforcement learning environments. Liam Fedus is the co-founder of Periodic Labs. At OpenAI, he was one of the researchers who helped create ChatGPT and served as VP of research leading post-training. In this talk, he shares how Periodic's AI explained experimental data that a landmark autonomous-lab study could not, surfacing a crystal structure found in no paper or database. He walks through…

    Watch on YouTube ↗

  • AI for Atoms: How Periodic Labs is Revolutionizing Materials Engineering with Co-Founder Liam Fedus

    April 03, 2026 · 📺 No Priors: AI, Machine Learning, Tech, & Startups

    What happens when you apply the scaling laws of large language models to the physical work of atoms? Elad Gil sits down with Liam Fedus, co-founder at Periodic Labs, which is pioneering an AI foundation lab for atoms. Liam discusses how he pivoted from dark matter physics research to the front lines of artificial intelligence, including stints at Google Brain and working on ChatGPT at OpenAI. He talks about how Periodic is connecting massive language models to the physical world to overcome data bottlenecks in material science. Liam also shares how they use language models as an orchestration…

    Watch on YouTube ↗

  • Will AI Replace Scientists? | Lifting with a ChatGPT Co-Creator

    April 01, 2026 · 📺 Swole as a Service

    Liam Fedus is the co-creator of ChatGPT and one of the founders of Periodic Labs, an AI scientific research startup that raised an enormous $300 million seed round. But today, Liam is in the house of pain (and gains!) as he comes to Swole as a Service to put his mind and body under pressure. Jayden puts Liam through a grueling (but satisfying!) arm workout, while quizzing him about whether AI will replace scientists in our near future. See how he does in our infamous Chestmaxxer Challenge push-up contest at the end… 00:21 Who is Liam Fedus? 01:21 Lying Skull Crushers & will AI replace scien…

    Watch on YouTube ↗

  • GPU Mode Keynote | Liam Fedus | Periodic Labs | How to Create an AI Scientist

    January 06, 2026 · 📺 Accel

    At Accel’s GPU Mode in San Francisco, Periodic Labs co-founder Liam Fedus made a case for accelerating science as the best possible use of our GPUs—and explained what’s standing in the way. To deal with noisy, messy, and incomplete data, he says, AI should return to the scientific playbook we’ve followed for centuries: conducting experiments. This keynote lays out why letting models into the lab is, in Liam’s view, the best way to unlock major leaps in discovery. #vc #venturecapital #founder #entrepreneurs #tech #CEO #liamfedus #periodiclabs #gpumode #gpu #ai #aiscience Accel: Accel is a gl…

    Watch on YouTube ↗

  • Periodic Labs: Training AI Scientists, with Liam Fedus & Ekin Dogus Cubuk (from a16z)

    October 02, 2025 · 📺 Cognitive Revolution "How AI Changes Everything"

    Today’s special crosspost features ‪@a16z‬ General Partner Anjney Midha with Liam Fedus and Ekin Dogus Cubuk of Periodic Labs. They discuss experiments in training frontier AI models on real-world physics lab experiments—rather than digital reward functions—and how coupling language models with automated materials synthesis could accelerate the discovery of room-temperature superconductors and transform R&D in advanced manufacturing. Shownotes brought to you by Notion AI Meeting Notes - try one month for free at: https://notion.com/lp/nathan -Core innovation: Periodic Labs is creating a "phy…

    Watch on YouTube ↗

  • AI for Discovery

    September 25, 2025 · 📺 Imagination in Action

    In this conversation, Emerson Collective investor Julian Barber interviews Liam Fedus, whose path runs from MIT dark-matter experiments to Google Brain, co-creating ChatGPT, and now founding Periodic to build a “foundation model for atoms.” Liam argues that accelerating discovery requires experiment-in-the-loop systems: AI agents that not only simulate but also run and learn from high-throughput, automated lab work to collapse hypothesis trees and create new knowledge. Starting in materials science—where iteration is fast and simulators are strong—Periodic aims to help industry intentionally d…

    Watch on YouTube ↗

  • AI Native 2025 Fireside Chat | Periodic Lab Co-Founder & Co- CEO, Liam Fedus

    September 25, 2025 · 📺 Zetta Venture Partners

    AI Native 2025 Fireside Chat | Periodic Lab Co-Founder & Co- CEO, Liam Fedus Follow Liam Fedus on LinkedIn   / william-liam-fedus-26547811   Follow Mark Gorenberg on LinkedIn   / markgorenberg   About AI Native Zetta Venture Partners launched AI Native with a clear vision: to create the defining forum where researchers, founders, and practitioners come together to shape the future of artificial intelligence. Unlike generic conferences, AI Native is rooted in the conviction that the next generation of breakthrough companies will be AI-native at their core—businesses built with AI as the fun…

    Watch on YouTube ↗

  • Liam Fedus & Barret Zoph - AI scaling with mixture of expert models

    April 20, 2022 · 📺 Towards Data Science

    Liam Fedus and Barrett Zoph, research scientists at Google Brain, came on the podcast to talk about AI scaling, sparsity and the present and future of MoE models. Intro music: ➞ Artist: Ron Gelinas ➞ Track Title: Daybreak Chill Blend (original mix) ➞ Link to Track:    • Ron Gelinas - Daybreak Chill Blend [ROYALT...   0:00 Intro 2:15 Guests’ backgrounds 8:00 Understanding specialization 13:45 Speculations for the future 21:45 Switch transformer versus dense net 27:30 More interpretable models 33:30 Assumptions and biology 39:15 Wrap-up

    Watch on YouTube ↗

The summary and quote tags on this profile are produced with AI assistance from verified, first-person interview transcripts, then checked by our team to confirm the speaker's identity and the accuracy of every quote. See how we verify →